Remaining Useful Life Prediction of Aircraft Engine Based on Bi-LSTM Network Integrated With Attention Mechanism

Abstract Predicting the remaining useful life (RUL) of an aircraft engine is crucial for ensuring the reliability and safety of an aircraft. This study has developed a novel data-driven hybrid network combining a Bidirectional Long Short-Term Memory (BiLSTM) with an attention mechanism to predict the RUL in aircraft engines. The model introduces an innovative approach by incorporating the feature-capture attention mechanism before the BiLSTM layer, which enhances the model’s focus on relevant sensor data segments to enable more effective feature extraction from multi-sensor data. This integration significantly enhances prediction accuracy compared to traditional shallow and deep learning models by leveraging the BiLSTM’s capability to analyze time-series data in both forward and backward directions. The proposed model demonstrates superior accuracy through comparative experiments conducted on the NASA C-MAPSS dataset, underscoring its potential to advance malfunction prediction and health management in aircraft engines.